RE: revisions and change in skos

Hi, 

I agree concepts do not die.

In some historic applications though you want to use the concept scheme
as a navigation structure.
So to know what concepts are applicable at a given time, you create
versions of them as I explained.

So indeed you want to know old and depricated concepts (and labels).
However, the application requirement we have is to be able to provide a
concept scheme as it was on a specific time.

I do not consider this to be in contradiction with the standard.

Hope this clarifies our use.

kr, Johan De Smedt. 
===================
-----Original Message-----
From: public-esw-thes-request@w3.org
[mailto:public-esw-thes-request@w3.org] On Behalf Of Aida Slavic
Sent: Monday, 29 September, 2008 11:14
To: De Smedt Johan
Cc: Rob Tice; public-esw-thes@w3.org
Subject: Re: revisions and change in skos


Johan,

What is a 'version concept'? I am not sure is it only a casual use of
the expression 'concept' instead of 'term' that causes problem - but it
may be worth to be more precise...

 > - URI are forever
 > - the skos:Concept may be constraind in time using an applicability
> period  > - the skos Concept has a creation date. modification date
and  > version(=introduction version) property  > - semantic relations
are not versioned (skos would be more difficult to  > accommodate that)
> - Next to semantic relations, change-notes are used on versioned  >
concepts.
 >   These change notes contain references to earlier/newr versions of a
 > concept

concept is an idea - this is the level SKOS is supposed to capture by
assigning an URI to this idea. New concepts can emerge - semantic field
of on concept can split into to or two concepts can be merged to create
a new concept. But in principle concepts themselves don't die,
disappear, change - so in principle URI is forever.
For SKOS two concepts from two different KOS schemes are always two
different concepts between which one can state the level of similarity.
To what degree two vocabularies speaks of the same concepts can be
stated through alignment/mapping
- and is defined by the mapping relationship. But even when we state
that two concepts means the 'same/equivalent' what is meant is that
their semantic fields overlap sufficiently for us to safely operate with
them as if they are the same.

Labels are sphere of practical use. Relationship between concepts and
terms
(labels) is  one to many. In real life have to add constraints to the
use of labels (preferred term) or change label for practical reasons.
Labels (i.e.
relationship between concept and label) can become depricated within
certain sphere of use and in certain time - but this has nothing to do
with the concept itself.

To make it more complicated - in principle in information indexing and
retrieval we need to preserve "the knowledge and the history of
label/concept relationship" - simply because older recorded documents
may have these labels used in the meaning that was relevant for the time
when document was created.
Good example are geographical and geo-political entities, historical
entities such as countries that don't exist anymore, political entities
that existed in certain time on which we still have documents. Then also
we need to know old and depricated term(s) for entities we now call
differently in order to be able to retrieve information on them from the
period when they were called differently.

I wonder whether my understanding here is correct - but I certainly
don't want to distract Rob and others trying to a good job here.

rgds

Aida

Received on Monday, 29 September 2008 09:27:14 UTC